@elonmusk@slashdot How great would it be to see a huge market overcorrection in the next N years, where so many fabs are churning out so much silicon that consumer prices reach all-time lows
The #AlphaFold 2 papers on the methods and human proteome predictions are out today in hard copy in @Nature! A really proud moment to see our work featured with a fantastic image on the front cover of the issue: https://t.co/tqKc7xmj98
The universal law of robustness is a tentative theoretical justification for *large* overparametrization in neural network learning. Here is a video explaining the law, in the context of other recent results on overparametrization (e.g., double descent).
https://t.co/LwfCN9pbsw
@ylecun I feel like Barlow Twins can be easily modified for BCI applications. Show two subjects the same stimuli, and use each of their brain signals as inputs to their respective arm of the model. Augmented images being analogous to the neural activity of two different individuals.